Publications list
Preprint
Scalable and Robust Multiband Modeling of AGN Light Curves in Rubin-LSST
Posted to a preprint site 26 Nov 2025
ArXiv.org
The Vera C. Rubin Observatory's Legacy Survey of Space and Time (LSST) will monitor tens of millions of active galactic nuclei (AGNs) for a period of 10 years with an average cadence of 3 days in six broad photometric bands. This unprecedented dataset will enable robust characterizations of AGN UV/optical variability across a wide range of AGN physical properties. However, existing tools for modeling AGN light curves are not yet capable of fully leveraging the volume, cadence, and multiband nature of LSST data. We present EzTaoX, a scalable light curve modeling tool designed to take advantage of LSST's multiband observations to simultaneously characterize AGN UV/optical stochastic variability and measure interband time delays. EzTaoX achieves a speed increase of$\sim 10^2-10^4 \times$on CPUs over current tools with similar capabilities, while maintaining equal or better accuracy in recovering simulated variability properties. This performance gain enables continuum time-delay measurements for all AGNs discovered by LSST -- both in the Wide Fast Deep survey and the Deep Drilling Fields -- thereby opening new opportunities to probe AGN accretion-flow geometries. In addition, EzTaoX's multiband capability allows robust characterization of AGN stochastic variability down to hourly timescales, facilitating the identification of accreting low-mass AGNs -- such as those residing in dwarf galaxies -- through their distinctive variability signatures.
Preprint
Posted to a preprint site 16 Aug 2025
A damped random walk (DRW) process is often used to describe the temporal UV/optical continuum variability of active galactic nuclei (AGN). However, recent investigations have shown that this model fails to capture the full spectrum of AGN variability. In this work, we model the 22-year-long light curves of $21,767$ quasars, spanning the redshift range $0.28 < z < 2.71$, as a noise-driven damped harmonic oscillator (DHO) process. The light curves, in the optical $g$ and $r$ bands, are collected and combined from the Sloan Digital Sky Survey, the Panoramic Survey Telescope and Rapid Response System, and the Zwicky Transient Facility. A DHO process can be defined using four parameters, two for describing its long-term behavior/variability, and the other two for describing its short-term behavior/variability. We find that the best-fit DHO model describes the observed variability of our quasar light curves better than the best-fit DRW model. Furthermore, the best-fit DHO parameters exhibit correlations with the rest-frame wavelength, the Eddington ratio, and the black hole mass of our quasars. Based on the power spectral density shape of the best-fit DHOs and these correlations, we suggest that the observed long-term variability of our quasars can be best explained by accretion rate or thermal fluctuations originating from the accretion disk, and the observed short-term variability can be best explained by reprocessing of X-ray variability originating from the corona. The additional information revealed by DHO modeling emphasizes the need to go beyond DRW when analyzing AGN light curves delivered by next-generation wide-field time-domain surveys.
Preprint
Quantifying the Active Galactic Nuclei Fraction in Cosmic Voids via Mid-Infrared Variability
Posted to a preprint site 15 Jul 2025
Observations and theoretical simulations suggest that the large scale environment plays a significant role in how galaxies form and evolve and, in particular, whether and when galaxies host an actively accreting supermassive black hole in their center (i.e., an Active Galactic Nucleus, or AGN). One signature of AGN activity is luminosity variability, which appears in the mid-infrared (mid-IR) when circumnuclear dust reprocesses UV and optical photons from the AGN accretion disk. We present here a suite of constraints on the fraction of AGN activity in the most underdense regions of the universe (cosmic voids) relative to the rest of the universe (cosmic walls) by using ~12 years of combined multi-epoch data from AllWISE and NEOWISE to quantify mid-IR variability. We find clear evidence for a larger mid-IR variability-AGN fraction among high and moderate-luminosity void galaxies compared to their wall counterparts. We also show that mid-IR variability identifies a rather large and unique population of AGNs, the majority of which have eluded detection using more traditional AGN-selection methods such as single-epoch mid-IR color selection. The fraction of these newly-recovered AGNs is larger among galaxies in voids, suggesting once again more prolific AGN activity in the most underdense large scale structures of the universe.
Preprint
HST /COS Lyman-alpha Absorbers in Cosmic Voids
Posted to a preprint site 14 Apr 2022
arXiv (Cornell University)
We investigate the spatial distribution of Lyman-$\alpha$ (Ly $\alpha$) absorbers within cosmic voids. We create a catalogue of cosmic voids in Sloan Digital Sky Survey Data Release 7 (SDSS DR7) with the VoidFinder algorithm of the Void Analysis Software Toolkit (VAST). Using the largest catalogue of low-redshift (z $\leq$ 0.75) IGM absorbers to date, we identify 392 Ly $\alpha$ absorbers inside voids. The fraction of Ly $\alpha$ absorbers inside voids (65 per cent) is comparable to the volume filling fraction of voids (68 per cent), and significantly greater than the fraction of galaxies inside voids (21 per cent). Inside voids, the spatial distribution of Ly $\alpha$ absorbers differs markedly from that of galaxies. Galaxy density rises sharply near void edges, while Ly $\alpha$ absorber density is relatively uniform. The radial distribution of Ly $\alpha$ absorbers inside voids differs marginally from a random distribution. We find that lower column density Ly $\alpha$ absorbers are more centrally concentrated inside voids than higher column density Ly $\alpha$ absorbers. These results suggest the presence of two populations of Ly $\alpha$ absorbers: low column density systems that are nearly uniformly distributed in the interiors of voids and systems associated with galaxies at the edges of voids.
Preprint
Posted to a preprint site 28 Dec 2020
ArXiv.org
The $\textit{Kepler}$ satellite potentially provides the highest precision
photometry of active galactic nuclei (AGN) available to investigate
short-timescale optical variability. We targeted quasars from the Sloan Digital
Sky Survey that lie in the fields of view of the $\textit{Kepler/K2}$
campaigns. Based on those observations, we report the discovery and properties
of a previously unidentified instrumental signature in K2. Systematic errors in
K2, beyond those due to the motion of the detector, plague our AGN and other
faint-target, guest-observer science proposals. Weakly illuminated pixels are
dominated by low frequency trends that are both non-astrophysical and
correlated from object to object. A critical clue to understanding this
instrumental noise is that different targets observed in the same channels of
Campaign 8 (rear facing) and Campaign 16 (forward facing) had nearly identical
light curves after time reversal of one of the campaigns. This observation
strongly suggests that the underlying problem relates to the relative
Sun-spacecraft-field orientation, which was approximately the same on day 1 of
Campaign 8 as the last day of Campaign 16. Furthermore, we measure that the
instrumental signature lags in time as a function of radius from the center of
the detector, crossing channel boundaries. Systematics documented in this
investigation are unlikely to be due to Moir\'{e} noise, rolling band, or
pointing jitter. Instead this work strongly suggests temperature-dependent
focus changes that are further subject to channel variations. Further
characterization of this signature is crucial for rehabilitating K2 data for
use in investigations of AGN light curves.
Preprint
AGN Variability Analysis Handbook
Posted to a preprint site 31 Oct 2018
arXiv.org
This work develops application techniques for stochastic modelling of Active
Galactic Nuclei (AGN) variability as a probe of accretion disk physics.
Stochastic models, specifically Continuous Auto-Regressive Moving Average
(CARMA) models, characterize lightcurves by estimating delay timescales that
describe movements away from and toward equilibrium (mean flux) as well as an
amplitude and frequency of intrinsic perturbations to the AGN flux. We begin
this tutorial by reviewing discrete auto-regressive (AR) and moving-average
(MA) processes, we bridge these components to their continuous analogs, and
lastly we investigate the significance of timescales from direct stochastic
modelling of a lightcurve projected in power spectrum (PSD) and structure
function (SF) space. We determine that higher order CARMA models, for example
the Damped Harmonic Oscillator (DHO or CARMA(2,1)) are more sensitive to
deviations from a single-slope power-law description of AGN variability; unlike
Damped Random Walks (DRW or CAR(1)) where the PSD slope is fixed, the DHO slope
is not. Higher complexity stochastic models than the DRW capture additional
covariance in data and output additional characteristic timescales that probe
the driving mechanisms of variability.
Preprint
Do the Kepler AGN Light Curves Need Re-processing?
Posted to a preprint site 15 Jul 2015
arXiv.org
Monthly Notices of the Royal Astronomical Society 2015 453 (2):
2075-2081 We gauge the impact of spacecraft-induced effects on the inferred variability
properties of the light curve of the Seyfert 1 AGN Zw 229-15 observed by
\Kepler. We compare the light curve of Zw 229-15 obtained from the Kepler MAST
database with a re-processed light curve constructed from raw pixel data
(Williams & Carini, 2015). We use the first-order structure function,
$SF(\delta t)$, to fit both light curves to the damped power-law PSD of
Kasliwal, Vogeley & Richards, 2015. On short timescales, we find a steeper
log-PSD slope ($\gamma = 2.90$ to within $10$ percent) for the re-processed
light curve as compared to the light curve found on MAST ($\gamma = 2.65$ to
within $10$ percent)---both inconsistent with a damped random walk which
requires $\gamma = 2$. The log-PSD slope inferred for the re-processed light
curve is consistent with previous results (Carini & Ryle, 2012, Williams &
Carini, 2015) that study the same re-processed light curve. The turnover
timescale is almost identical for both light curves ($27.1$ and $27.5$~d for
the reprocessed and MAST database light curves). Based on the obvious visual
difference between the two versions of the light curve and on the PSD model
fits, we conclude that there remain significant levels of spacecraft-induced
effects in the standard pipeline reduction of the Kepler data. Re-processing
the light curves will change the model inferenced from the data but is unlikely
to change the overall scientific conclusion reached by Kasliwal et al.
2015---not all AGN light curves are consistent with the DRW.
Preprint
Posted to a preprint site 26 Aug 2014
arXiv.org
Monthly Notices of the Royal Astronomical Society 2014, Volume 444, Issue 2: p.3559-3570 We measure the HI mass function (HIMF) and velocity width function (WF) across environments over a range of masses $7.2<\log(M_{HI}/M_{\odot})<10.8$, and profile widths $1.3\log(km/s)<\log(W)<2.9\log(km/s)$, using a catalog of ~7,300 HI-selected galaxies from the ALFALFA Survey, located in the region of sky where ALFALFA and SDSS (Data Release 7) North overlap. We divide our galaxy sample into those that reside in large-scale voids (void galaxies) and those that live in denser regions (wall galaxies). We find the void HIMF to be well fit by a Schechter function with normalization $\Phi^*=(1.37\pm0.1)\times10^{-2} h^3Mpc^{-3}$, characteristic mass $\log(M^*/M_{\odot})+2\log h_{70}=9.86\pm0.02$, and low-mass-end slope $\alpha=-1.29\pm0.02$. Similarly, for wall galaxies, we find best-fitting parameters $\Phi^*=(1.82\pm0.03)\times10^{-2} h^3Mpc^{-3}$, $\log(M^*/M_{\odot})+2\log h_{70}=10.00\pm0.01$, and $\alpha=-1.35\pm0.01$. We conclude that void galaxies typically have slightly lower HI masses than their non-void counterparts, which is in agreement with the dark matter halo mass function shift in voids assuming a simple relationship between DM mass and HI mass. We also find that the low-mass slope of the void HIMF is similar to that of the wall HIMF suggesting that there is either no excess of low-mass galaxies in voids or there is an abundance of intermediate HI mass galaxies. We fit a modified Schechter function to the ALFALFA void WF and determine its best-fitting parameters to be $\Phi^*=0.21\pm0.1 h^3Mpc^{-3}$, $\log(W^*)=2.13\pm0.3$, $\alpha=0.52\pm0.5$ and high-width slope $\beta=1.3\pm0.4$. For wall galaxies, the WF parameters are: $\Phi^*=0.022\pm0.009 h^3Mpc^{-3}$, $\log(W^*)=2.62\pm0.5$, $\alpha=-0.64\pm0.2$ and $\beta=3.58\pm1.5$. Because of large uncertainties on the void and wall width functions, we cannot conclude whether the WF is dependent on the environment.
Preprint
Photometric Properties of Void Galaxies in the Sloan Digital Sky Survey DR7 Data Release
Posted to a preprint site 08 May 2012
arXiv.org
Using the sample presented in Pan:2011, we analyse the photometric properties of 88,794 void galaxies and compare them to galaxies in higher density environments with the same absolute magnitude distribution. In Pan et al. (2011), we found a total of 1054 dynamically distinct voids in the SDSS with radius larger than 10h^-1 Mpc. The voids are underdense, with delta rho/rho < -0.9 in their centers. Here we study the photometric properties of these void galaxies. We look at the u - r colours as an indication of star formation activity and the inverse concentration index as an indication of galaxy type. We find that void galaxies are statistically bluer than galaxies found in higher density environments with the same magnitude distribution. We examine the colours of the galaxies as a function of magnitude, and we fit each colour distribution with a double-Gaussian model for the red and blue subpopulations. As we move from bright to dwarf galaxies, the population of red galaxies steadily decreases and the fraction of blue galaxies increases in both voids and walls, however the fraction of blue galaxies in the voids is always higher and bluer than in the walls. We also split the void and wall galaxies into samples depending on galaxy type. We find that late type void galaxies are bluer than late type wall galaxies and the same holds for early galaxies. We also find that early type, dwarf void galaxies are blue in colour. We also study the properties of void galaxies as a function of their distance from the center of the void. We find very little variation in the properties, such as magnitude, colour and type, of void galaxies as a function of their location in the void. The only exception is that the dwarf void galaxies may live closer to the center. The centers of voids have very similar density contrast and hence all void galaxies live in very similar density environments (ABRIDGED)
Preprint
Topology of structure in the Sloan Digital Sky Survey: model testing
Posted to a preprint site 25 Oct 2006
ArXiv.org
Astrophys.J.675:16-28,2008 We measure the three-dimensional topology of large-scale structure in the
Sloan Digital Sky Survey (SDSS). This allows the genus statistic to be measured
with unprecedented statistical accuracy. The sample size is now sufficiently
large to allow the topology to be an important tool for testing galaxy
formation models. For comparison, we make mock SDSS samples using several
state-of-the-art N-body simulations: the Millennium run of Springel et al.
(2005)(10 billion particles), Kim & Park (2006) CDM models (1.1 billion
particles), and Cen & Ostriker (2006) hydrodynamic code models (8.6 billion
cell hydro mesh). Each of these simulations uses a different method for
modeling galaxy formation. The SDSS data show a genus curve that is broadly
characteristic of that produced by Gaussian random phase initial conditions.
Thus the data strongly support the standard model of inflation where Gaussian
random phase initial conditions are produced by random quantum fluctuations in
the early universe. But on top of this general shape there are measurable
differences produced by non-linear gravitational effects (cf. Matsubara 1994),
and biasing connected with galaxy formation. The N-body simulations have been
tuned to reproduce the power spectrum and multiplicity function but not
topology, so topology is an acid test for these models. The data show a
``meatball'' shift (only partly due to the Sloan Great Wall of Galaxies; this
shift also appears in a sub-sample not containing the Wall) which differs at
the 2.5\sigma level from the results of the Millennium run and the Kim & Park
dark halo models, even including the effects of cosmic variance.